20 - Data transformations

Question 1

Identifying a lack-of-fit to a line

Answer 1

• residuals plots are key to identifying a systematic lack-of-fit
• smooths of a scatterplot are helpful too

Question 2

Scatterplot smoother

Answer 2

a data-driven summary of the association between 2 variables

can pick up the curvature in the data and suggest a non-linear model

Question 3

On observing curvature, _______

Answer 3

transform

• a data transformation takes a column in a spreadsheet, applies a function to each row in the column, and creates a new transformed column

Question 4

Why can we transform X with little concern?

Answer 4

bc its role is a predictor

when we model, we don’t treat X and Y as equal partners:

Y is the focus of the exercise

*be very careful about transforming Y*